Laminar is an open-source tool designed for observability and analytics in complex LLM applications. It combines the functionality of DataDog and PostHog for LLM systems, offering automatic instrumentation for vector DB calls and function tracking. Background job queues are processed to create metrics, allowing users to design custom pipelines for specific data extraction. Laminar is built for scalability, with a modern stack including Rust, RabbitMQ, Postgres, and Clickhouse. Unique features include semantic events-based analytics, insightful dashboards, and the ability to create and manage prompt chain pipelines through the UI. Laminar offers easy deployment options, such as a managed platform or self-hosting with Docker compose.
https://github.com/lmnr-ai/lmnr